fAIr Models¶
Model registry and ML pipeline orchestration for fAIr.
fair-py-ops is the Python package model developers install to build
ZenML pipelines, validate
STAC items, and test the full workflow locally
before submitting to the fAIr platform.
What is fAIr?¶
Humanitarian AI for OpenStreetMap
fAIr is a humanitarian AI platform by HOT that enables feature extraction from very high resolution aerial imagery (buildings, roads, trees) for OpenStreetMap mapping. This repository provides the ML infrastructure layer.
Key Concepts¶
| Concept | Description |
|---|---|
| Base model | A reusable ML blueprint contributed via PR (e.g. UNet, YOLOv8) |
| Local model | A finetuned model produced by a ZenML pipeline on user data |
| STAC catalog | The model/dataset registry using MLM and Version extensions |
| ZenML pipeline | Orchestrated training and inference workflows |
Quick Links¶
Start here
- Getting Started : install, run example, understand the workflow
- Architecture : STAC catalog structure, flows, infrastructure
- Contributing a Model : add your model to fAIr
- API Reference : Python package documentation
- Changelog : release history